Skyview processing of abnormal ADS-B input

[BrentC48]

Purpose of this post is to point out a serious problem with Skyview/autopilot system when erroneous ADS-B input is entered into Skyview.  Earlier today I was flying near Carroll County airport MD (KDMW) and pressed the Baro button to have the nearest Baro setting entered into Skyview via the Dynon 470 unit.  For whatever reason, as you can see the local Baro setting was clearly incorrect, but Skyview accepted the odd value and displayed my altitude very incorrectly as a result.  As I was flying via autopilot, the aircraft immediately began to dive which was obviously not my intent.  Reset the baro setting and tried this a second time, but again Skyview accepted the very odd value from Carroll Cty.  Screenshots are from second attempt.  Very odd.

While I don't understand why the electronically communicated ADS-B Baro setting from this airport was so incorrect, it would be ideal if Skyview had limits or filters to avoid such incorrect values from automatically being accepted into the system. 

 

[KRviator]

We don't have ADS-B weather down here, but presumably, there is no way to tell you're about to uplink such a faulty baro setting until you actually note it on the subscale, or look at it in the APT-Wx page beforehand?

And that was 2700 knob clicks to get back from 3.01 to 30.18 each time?!?

 

[swatson999]

An interesting error, albeit one in the ADS-B data rather than SV...

I know there's an FAR that covers non-operation if the baro is 21" or higher, but I am unaware of one for *low* settings. It does seem like a reasonable use case to have the user confirm entry of an abnormally low or high value (say, something less than 27" or higher than 31"?).

Bet that got your attention!

 

[swatson999]

BTW, that looks all kinds of messed up...like the altitude wrapped to a negative number, but you say the autopilot started to *descend*? Am I seeing that correclty?

 

[BrentC48]

BTW, that looks all kinds of messed up...like the altitude wrapped to a negative number, but you say the autopilot started to *descend*?  Am I seeing that correclty?

Good question, everything happened so fast and I disconnected the AP before totally getting out of control.  My "impression" was that the AP was putting me into a dive, but logically it should have started a climb.  Maybe such a dramatic Baro driven alt. change also messes up the AP.  In anycase, it was about to make an unexpected and significant altitude change which is my bigger concern. 

Totally agree that the core issue is with the ADS-B feed itself, but ideally Skyview could help address this with some filters on Baro changes that are more than "X" units without confirming that with the pilot. Overall, I really like the convenient quick Baro update feature.

 

[Dynon]

Interesting find there. I'll make sure to report it to our development team so that they can take a closer look at it. There are definitely some limits checking that we can do to prevent erroneous data from adversely affecting SkyView here.

 

[swatson999]

Interesting find there. I'll make sure to report it to our development team so that they can take a closer look at it. There are definitely some limits checking that we can do to prevent erroneous data from adversely affecting SkyView here.

Not to put too pedantic a point on it, but I'll repeat what I tell all of my software teams...never, ever trust incoming data to any function, routine, etc., to be correct. Always assume that anything that *can* happen to the data, *will* happen to it

And yet, they still ignore my sage advice at times...

Yeah, especially with data like this that has so many links in the chain, from the observer or sensor, through the local systems, then to the nationwide systems, then via RF, then into the airborne systems...lots of places for it to go awry. It's actually impressive that it doesn't more often...

 

[Raymo]

I would suggest also reporting this to the appropriate FAA and/or Carroll County authority so they can check their system.

 

[dynonsupport]

It does appear that the airport reported the baro wrong, as you can see the incorrect values in the METAR history:

https://www.wunderground.com/histor...eqdb.zip=21158&reqdb.magic=55&reqdb.wmo=99999

 

[swatson999]

Not just once, either...several times throughout the day...

Wonder what other data is wrong that gets sent out that there's no way to know?